Attractor Dynamics in Realistic Hippocampal Networks

Elliot D. Menschik, Shih-Cheng Yen, and Leif H. Finkel

Department of Bioengineering
3320 Smith Walk, 301 Hayden Hall
University of Pennsylvania
Philadelphia, PA 19104, U. S. A.

Abstract

Attractor-based networks offer a powerful paradigm for the storage and recall of memory.  However, as traditionally formulated (Hopfield, 1982; Amit, 1989) with abstracted neurons, these networks bear only an indirect relation to actual hippocampal and neocortical networks. Here we present an attractor network instantiated in a detailed biophysical-level model of the CA3 region of the hippocampus, simulated using GENESIS. The network is comprised of the 66-compartment pyramidal cells and 51-compartment interneurons of Traub and colleagues (Traub et al., 1994; Traub and Miles, 1995) and takes advantage of the mechanism of pyramidal cell synchrony imposed by networks of interneurons oscillating in the gamma range (Traub et al., 1996b; Wang and Buzsaki, 1996; Bush and Sejnowski, 1996).  We demonstrate that such a network rapidly converges to stored memory states (metastable attractors) and has interesting temporal dynamics that arise from the interplay of AMPA, NMDA, and GABA_A synapses (particularly as influenced by neuromodulators) and our studies suggest that a cellular-level model of CA3 can exhibit functional-level properties.

[ Full Text in PDF ]